Article Text

Original research
Cost-effectiveness of strength exercise or aerobic exercise compared with usual care for patients with knee osteoarthritis: secondary results from a multiarm randomised controlled trial in Norway
  1. Rikke Munk Killingmo1,2,
  2. Britt Elin Øiestad1,
  3. May-Arna Risberg2,3,
  4. Esther Maas4,5,
  5. Margreth Grotle1,6
  1. 1Department of Rehabilitation Science and Health Technology, Oslo Metropolitan University, Oslo, Norway
  2. 2Division of Orthopaedic Surgery, Oslo Universitetssykehus, Oslo, Norway
  3. 3Department of Sports Medicine, Norwegian School of Sports Sciences, Oslo, Norway
  4. 4Department of Health Sciences, Vrije University Amsterdam, Amsterdam, The Netherlands
  5. 5The Amsterdam Movement Sciences Research Institute, Amsterdam, The Netherlands
  6. 6Research and Communication Unit for Musculoskeletal Health, Oslo universitetssykehus Ulleval, Oslo, Norway
  1. Correspondence to Dr Rikke Munk Killingmo; rikke{at}oslomet.no

Abstract

Objectives To evaluate the 1-year cost-effectiveness of strength exercise or aerobic exercise compared with usual care for patients with symptomatic knee osteoarthritis (OA), from a societal and healthcare perspective.

Design Cost-effectiveness analysis embedded in a three-arm randomised controlled trial.

Participants and setting A total of 161 people with symptomatic knee OA seeking Norwegian primary or secondary care were included in the analyses.

Interventions Participants were randomised to either 12 weeks of strength exercise (n=54), 12 weeks of aerobic exercise (n=53) or usual care (n=54).

Outcome measures Quality-adjusted life-years (QALYs) estimated by the EuroQol-5 Dimensions-5 Levels, and costs related to healthcare utilisation and productivity loss estimated in euros (€), aggregated for 1 year of follow-up. Cost-effectiveness was expressed with mean incremental cost-effectiveness ratios (ICERs). Bootstrapping was used to estimate ICER uncertainty.

Results From a 1-year societal perspective, the mean cost per patient was €7954, €8101 and €17 398 in the strength exercise, aerobic exercise and usual care group, respectively. From a 1-year healthcare perspective, the mean cost per patient was €848, €2003 and €1654 in the strength exercise, aerobic exercise and usual care group, respectively. Mean differences in costs significantly favoured strength exercise and aerobic exercise from a 1-year societal perspective and strength exercise from a 1-year healthcare perspective. There were no significant differences in mean QALYs between groups. From a 1-year societal perspective, at a willingness-to-pay threshold of €27 500, the probability of strength exercise or aerobic exercise being cost-effective was ≥98%. From a 1-year healthcare perspective, the probability of strength exercise or aerobic exercise being cost-effective was ≥97% and ≥76%, respectively.

Conclusion From a 1-year societal and healthcare perspective, a 12-week strength exercise or aerobic exercise programme is cost-effective compared with usual care in patients with symptomatic knee OA.

Trial registration number NCT01682980.

  • health economics
  • physical therapy modalities
  • primary health care

Data availability statement

Data are available on reasonable request. Data supporting findings of this study are not public available as participants have not consented for their data to be public available. However, data are available from the corresponding author on reasonable request and with permission of an ethics committee and the Oslo University Hospital (contact through corresponding author).

https://creativecommons.org/licenses/by/4.0/

This is an open access article distributed in accordance with the Creative Commons Attribution 4.0 Unported (CC BY 4.0) license, which permits others to copy, redistribute, remix, transform and build upon this work for any purpose, provided the original work is properly cited, a link to the licence is given, and indication of whether changes were made. See: https://creativecommons.org/licenses/by/4.0/.

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

STRENGTHS AND LIMITATIONS OF THIS STUDY

  • The main strength of the present study is that it was a cost-effectiveness study of a three-arm randomised controlled trial (RCT), that it was conducted in line with the Panel on Cost-Effectiveness in Health and Medicine guidelines, and preplanned with a published statistical analysis plan. Furthermore, that bootstrapping methods were used to determine uncertainty around calculated incremental cost-effectiveness ratios.

  • The main limitation of the present study is that it was powered on the primary outcome of the three-arm RCT, not on the cost-effectiveness outcome. However, this is common practice and bootstrapped estimates for result interpretation were used.

  • The present study excluded people >70 years of age, and it is unknown whether results are generalisable to this subgroup of people with knee osteoarthritis. Costs related to productivity loss are expected to be markedly lower among people >70 years. Moreover, people with self-reported body mass index >35 kg/m2 were excluded.

  • Due to between-country differences in healthcare organisations and mechanisms of compensation for productivity loss, readers are advised to be cautious with the generalisation of results to other healthcare systems.

Introduction

The burden of knee osteoarthritis (OA) has been growing along with an increasing and ageing population.1 2 Knee OA imposes a leading cause of disability globally,1–4 and an extensive burden to society with high costs related to both healthcare utilisation and productivity loss.1 4–6 Clinical guidelines for the management of knee OA consistently recommend patient education, exercise and weight management (if appropriate) as core treatments, regardless of disease severity and comorbidity.7–9 Core treatments have been stated to be safe, appropriate and effective, yet uptake of these proven treatments seems to be low in clinical practice.10 11 One approach to support implementation of core treatments in healthcare systems and reimbursement plans is to evaluate cost-effectiveness (CE).11–13 CE analyses (CEAs) explicitly quantify the relative costs and benefits of different interventions. Thus, CEAs aim to illuminate potential trade-offs and inform discussions on resource allocation in healthcare.6 12 The evidence base for CE of knee OA core treatments is still relatively limited,11 14 and researchers have highlighted that further knowledge is needed to bolster implementation of appropriate healthcare plans, improve the use of scarce healthcare resources and reduce the global burden of knee OA.6 11 15

A three-arm randomised controlled trial (RCT)16 evaluated the effectiveness of a 12-week strength exercise (SE) or aerobic exercise (AE) programme, compared with usual care (UC), on knee-related quality of life after 1 year in individuals with symptomatic knee OA. The trial found no statistically significant difference in effects between groups. A result generally in accordance with previous research17 showing a gradual decrease in exercise effects over time, being no better than UC at around 9 months. However, the lack of statistical differences in effects does not necessarily mean that interventions are identical in terms of CE. Therefore, the Professional Society for Health Economics and Outcomes Research Cost-Effectiveness Analysis Alongside Clinical Trials (RCT-CEA) Task Force18 recommend researchers to perform CEAs regardless of clinical results. Furthermore, it has been recommended to conduct these analyses preferably from both a societal and healthcare perspective.6 12 18–20

Therefore, the aim of this study was to evaluate the 1-year CE of SE or AE compared with UC for patients with symptomatic knee OA, from a societal and healthcare perspective.

Method

The method of this study has been reported previously in the study protocol21 22 and in the paper on clinical effectiveness.16 For the CE part of the study, the reference case for economic evaluations in OA,6 the Panel on Cost-Effectiveness in Health and Medicine guidelines,19 20 as well as the Consolidated Health Economic Evaluation Reporting Standards (CHEERS) Statement23 were used. A summary of the method is presented below.

Design and setting

This study is a CEA embedded in a three-arm RCT, evaluating the effect of SE or AE, compared with UC, on knee-related quality of life in individuals with symptomatic knee OA. The trial had a treatment phase of 2 plus 12 weeks and a follow-up phase of 12 months.

Participants, recruitment procedure and randomisation

Eligible participants were people with symptomatic knee OA (confirmed mild to moderate radiographic, Kellgren and Lawrence24 grades 2–3), aged 35–70 years, who fulfilled 3/4 of the American College of Rheumatology clinical criteria (stiffness <30 min, crepitus, osteophytes, pain last days last month).25 Excluded were those with serious physical or mental illness, serious musculoskeletal impairment in the lower extremities or low back, lower extremity prosthesis, self-reported body mass index >35 kg/m2, scheduled surgery within 6 months, Norwegian language barriers and those already participating in structured, weekly, SE or AE. Participants were recruited from primary and secondary care in Norway between April 2013 and March 2020 (the main reasons for a slow recruitment rate were a lack of time and few resources available to recruit on a weekly basis). All included participants signed an informed consent before study enrolment. Participants were randomised to either SE, AE or UC with a 1:1:1 ratio within block sizes of 6. A biostatistician, not involved in the project, prepared the computer-generated randomisation lists. Group allocation was concealed for the researchers and the self-reported data assessors.

Interventions

A detailed description of the rationale, development and content of the interventions can be found elsewhere.16 Briefly, the SE group was followed up closely by an experienced physiotherapist, received an individual, supervised programme based on a previously developed exercise programme for knee patients (including neuromuscular and SEs)26 and was told to exercise 2–3 times a week for 12 weeks plus a 2-week preparation phase to be familiar with the programme and optimise exercise techniques. The AE group received a stationary cycling programme based on guidelines for aerobic training parameters in people with pain associated with OA (10 min warm up, 30 min on moderate intensity (75% of maximum heart rate) and 5 min on low intensity)27 and was told to exercise 2–3 times a week for 12 weeks plus a 2-week preparation phase. The number of physiotherapy consultations included in the SE and AE intervention was flexible and individually determined. A minimum of two exercise sessions per week were supervised by the physiotherapist while the third session could be performed unsupervised at home. All supervised SE and AE exercise sessions took place at physiotherapy clinics situated near the participant’s homes. All participants exercised individually. The UC group was told to avoid starting a new exercise programme involving structured SE or AE until the 4 months of follow-up was completed but otherwise live as usual.

Data collection, effect and cost measures

At baseline (before random group allocation), all participants completed a comprehensive questionnaire. Follow-up questionnaires covering healthcare utilisation, productivity loss (sick leave, work assessment allowance and disability benefits) and health state were sent at 14 weeks (postintervention) and 12 months after inclusion. All questionnaires were completed on paper. In addition, healthcare utilisation and productivity loss were also recorded at 6 and 9 months after inclusion by phone interviews. Furthermore, public registry data on knee replacement surgery was collected from the Norwegian Arthroplasty Register, in the period from baseline to 12 months follow-up.

Effect measure

The effect measure of this study was health-related quality of life expressed by quality-adjusted life-years (QALYs). First, health state of the participants was measured by the EuroQol-5 Dimensions-5 Levels (EQ-5D-5L).28 Second, the UK tariff29 was used to convent health states into utility scores (range −0.59 to 1), anchored at 0 ‘death’ and 1 ‘perfect health’, with negative values representing health states perceived to be worse than death. The UK tariff was used, as a Norwegian tariff is not available. QALYs were estimated as area under the curve using the trapezoidal method.30 QALYs remain undiscounted due to follow-up being confined to 1 year.30

Cost measures

This study adopted a societal and healthcare perspective, thus both costs related to healthcare utilisation and productivity loss were included. Healthcare utilisation due to knee OA (including physiotherapist consultations related to the study interventions) was self-reported and included consultations with healthcare professionals (type and frequency) (online supplemental additional file 1, figure A1). In addition, information on knee replacement surgery was obtained from the Norwegian Arthroplasty Register. Productivity loss due to knee OA was self-reported and included the number of calendar days with sick leave, work assessment allowance and disability benefits (online supplemental additional file 1, figure A1). Workers in Norway qualify for sickness benefits from the public welfare agency if they have been in paid work for minimum 4 weeks before the sickness incident, and if the occupational disability is documented by a doctor’s sick leave certificate. In general, sickness benefits can be received from the first day of reported sick and up to 1 year. If a person is still unable to work after 1 year, he or she may be entitled to work assessment allowance or disability benefits.

Healthcare utilisation and productivity loss were reported with a 3-month recall period at each time point of follow-up. A number of calendar days with productivity loss were converted into workdays with productivity loss and adjusted for employment rate as well as reported grading of productivity loss, summarising the number of workdays with part-time productivity loss to the number of workdays with complete productivity loss.

Analyses

All analyses were preplanned in a statistical analysis plan published a priori31 and performed using the IBM SPSS V.26 (IBM) or Stata version 16.1 (StataCorp). The intention-to-treat (ITT) method was used. P values <0.05 were considered statistically significant. All statistical tests were two sided.

Study flow

Flow of participants through the study was reported according to the Consolidated Standards of Reporting Trials Statement32 with a flow chart. Baseline differences between responders (at all time points) and non-responders (at minimum one follow-up time point) were evaluated through frequency and percentage.

Missing data

Missing value patterns were visually explored, and missingness at random was assumed. Also, we found evidence against the hypothesis that values were not missing completely at random (Little’s test, p>0.05). Missing data were handled by multiple imputation. Five multiple imputation datasets with 10 iterations were created using regression estimation, and pooled estimates were calculated using Rubin’s rules.33

Healthcare utilisation, productivity loss and cost estimation

Type and frequency of use of different healthcare resources (including physiotherapist consultations related to the study interventions) and type and frequency of productivity loss were calculated for each of the follow-up periods. Costs of healthcare utilisation per patient were estimated by multiplying frequency of use by unit costs collected from national pricelists (table 1). Non-healthcare costs related to the provision of healthcare (such as transportation and exercise equipment) were not estimated. Costs of productivity loss per patient were estimated by multiplying number of workdays with complete productivity loss by an estimated average wage rate including taxes and social costs (table 1). All costs were presented in euros (€) 2020 and estimated with mean (SD) values for the three groups, for the following follow-up periods: 0 to 3 months, >3 to 12 months and 0 to 12 months. Costs remain undiscounted due to follow-up being confined to 1 year.30 Norwegian prices were recalculated to euros using the exchange rate from the Norwegian Bank of Norway from February 2020 (€1=NOK10).

Table 1

Cost categories, units, unit price, all numbers in euros (€) and Norwegian kroner (NOK) for 2020

CE analysis

Differences between groups in costs (SE or AE–UC) were described with means (95% CI) and evaluated with Student’s t-test. Differences between groups in QALYs (SE or AE–UC) were described with means (95% CI) and evaluated with ANCOVA to adjust for baseline scores unequally distributed across the three groups at baseline (sex, occupational status, EQ-5D-5L score and healthcare utilisation prior to inclusion). CE was estimated with mean incremental cost-effectiveness ratios (ICERs) by dividing mean differences in costs by mean differences in effects. The CE threshold (willingness-to-pay, WTP) for OA was based on the Norwegian governmental report no. 34 to the parliament with a value of €27 500 per QALY.34 To illustrate uncertainty surrounding the ICERs, bootstrapped (10 000 replicated datasets) cost and effect pairs were plotted on CE planes (CE planes) and CE acceptability curves (CEACs). A CE plane is divided into four quadrants: the northwest (NWQ), the northeast (NEQ), the southwest (SWQ) and the southeast (SEQ), representing all combinations of possible outcomes. If most incremental cost-effectiveness pairs (ICERs) are located in the NWQ, the intervention (SE or AE) is assumed to be more costly and less effective than the benchmark (UC). Whereas the NEQ indicates that the intervention is more costly and more effective than the benchmark, the SWQ that the intervention is less costly and less effective, and the SEQ that the intervention is less costly and more effective. The CEACs were used to demonstrate the probability that the intervention (SE or AE) is cost-effective in comparison to UC for a range of different WTP values.

Sensitivity analysis

To evaluate the credibility of the imputation procedure on missing values, two sensitivity analyses were performed on the main analysis: (1) complete-case analysis without adjustment for missing data and (2) without outliers. Outliers were identified with simple scatterplots by visual inspection and defined as patients with somewhat remarkably high total costs; 7 patients with costs ≥€20 772 at 0–3 months, and 13 patients with costs ≥€52 289 at 0–12 months. Of the outliers, 13 patients had costs related to primary care consultations, 11 to medical specialist consultations, 10 to sick leave, 6 to work assessment allowance, 4 to surgery and 3 to disability benefits. To evaluate the effect of not including transportation costs related to the interventions, a sensitivity analysis including an estimated average transportation cost was performed. Transportation costs per patient were estimated by multiplying the number of physiotherapy consultations in the intervention period by an estimated average transportation cost of €7 (based on a government-set rate of €0.35 per km for an estimated average round-trip distance of 20 km). To evaluate the appropriateness of analysing groups’ differences in costs (skewed data) with a parametric test (Student’s t-test), a sensitivity analysis was conducted with the Mann-Whitney U test. To evaluate the effect of single variables on the ICERs, a multiple one-way sensitivity analysis was performed. Relevant costs and QALYs were varied 20% below and above estimates used in the main analysis (total costs, 0–12 months), and Tornado diagrams were used to visually demonstrate the results.

Sample size

This study contains secondary analysis embedded in a three-arm RCT. Details on sample size calculation related the primary aim are provided elsewhere.16 21

Patient and public involvement

This study was designed when user involvement was not a mandatory part of trials. Yet, the SE programme was developed in collaboration with clinicians with high competence in treating patients with knee OA. Results will be disseminated to the recruiting primary and secondary care providers and the participating patients.

Results

A total of 168 patients were randomised into the three groups: 55 to SE, 56 to AE and 57 to UC. Seven participants (4%) withdrew their informed consent (due to lack of time or not attending tests), leaving 161 (96%) participants for the ITT analyses (54 to SE, 53 to AE and 54 to UC). Flow of participants through the study is shown in online supplemental additional file 2, figure A2. Table 2 shows patient characteristics and clinical status at baseline, along with the proportion of missing data per variable. Overall, the SE and AE groups were balanced as compared with the UC group, and there were no statistically significant differences in EQ-5D-5L scores between the SE or AE group as compared with the UC group. There were only minor differences between responders and non-responders (online supplemental additional file 3, table A1).

Table 2

Patient characteristics and clinical status at baseline

Missing data ranged from 0.0% to 3.0% for included baseline variables and 0.0% to 27.0% for follow-up variables (healthcare utilisation, productivity loss, EQ-5D-5L) used to calculate the outcome values.

Healthcare utilisation, productivity loss, cost estimation and effect

Table 3 shows healthcare utilisation and productivity loss throughout 1 year of follow-up for all participants. Online supplemental additional file 4, tables A2–A4 show healthcare utilisation (including physiotherapist consultations related to the study interventions) and productivity loss throughout 1 year of follow-up for the three groups separately.

Table 3

Healthcare utilisation and productivity loss throughout 1 year of follow-up (n=161)

Mean (SD) intervention cost per patient was €190 (€433) and €201 (€580) for the SE and the AE intervention, respectively. Costs related to healthcare utilisation (including physiotherapist consultations related to the study interventions) and productivity loss for the 1 year of follow-up are shown in table 4. Online supplemental additional file 5, table A5 shows costs related to healthcare utilisation and productivity loss for the follow-up periods 0 to 3 months and >3 to 12 months. Costs were mainly related to productivity loss, accounting for 75%, 89% and 91% of total costs during 1 year of follow-up in the SE, AE and UC group, respectively. During the 1 year of follow-up, the mean cost per patient related to healthcare utilisation was approximately half in the SE group as compared with the AE and UC group (€848 vs €2003 and €1654). Whereas the mean cost per patient related to productivity loss was approximately half in the SE and AE group as compared with the UC group (€7105 and €6098 vs €15 745). In total, during the 1 year of follow-up, the sum of all costs was statistically significant lower in the SE and AE groups as compared with the UC group, with an estimated mean (95% CI) difference per patient at €−9445 (€−17 599 to €−1291) and €−9297 (€−17 223 to €−1371), respectively (table 5). The sensitivity analysis conducted with the Mann-Whitney U test provided similar results in terms of statistical significance.

Table 4

Costs (€) due to healthcare utilisation and productivity loss throughout 1 year of follow-up for all treatment groups*

Table 5

Mean cost (€) and effect (QALYs) differences (95% CI) between the intervention and usual care group during the period 0–3 and 0–12 months, including ICER, cost-effectiveness plane distributions and sensitivity analysis

At all time points, there were no statistically significant difference in mean QALYs between the SE or AE group as compared with the UC group (table 5). The sensitivity analyses showed overall no substantial change in point estimates of mean cost (€) and effect (QALYs) differences when comparing complete case analysis, and analysis without outliers to the main analysis (table 5). Furthermore, no substantial change in point estimates of mean cost (€) and effect (QALYs) differences was seen when comparing analysis including transportation costs related to the intervention to the main analysis (online supplemental additional file 6, table A6).

Cost-effectiveness

Comparing SE to UC at 1 year of follow-up, all estimated ICERs were found to be with negative values, thus indicating that SE on average was less costly and more effective for improving QALYs than UC, both from a societal and a healthcare perspective. The CE planes showed similar results with most incremental CE pairs (86%–89%) being located in the SEQ (figure 1 and table 5). The CEACs indicated that the probability of SE being cost-effective compared with UC was ≥95% for all WTP thresholds.

Figure 1

Cost-effectiveness plane and cost-effectiveness acceptability curve for different ceiling ratios (EUR) for quality-adjusted life-years (QALYs) indicating the probability of cost-effectiveness of strength exercise or aerobic exercise vs usual care on total costs (healthcare utilisation and productivity loss) and healthcare utilisation costs at 0–12 months. All estimates are based on bootstrapping (10 000 replicated datasets). The dashed line represents the willingness to pay (WTP) threshold of €27 500. EUR, euros.

Comparing AE to UC at 1 year of follow-up, we found a greater variation in the results. From a societal perspective, the estimated ICERs were found to be with a negative value, thus indicating that AE on average was less costly and more effective for improving QALYs than UC. The CE plane showed likewise that most incremental CE pairs (85%) were located in the SEQ (figure 1 and table 5). Furthermore, the CEAC indicated that the probability of AE being cost-effective compared with UC was ≥98% for all WTP thresholds. From a healthcare perspective, however, the estimated ICER was €8725 per QALY gained. The CE plane showed that most incremental CE pairs (55%) were located in the NEQ (figure 1 and table 5), thus indicating that AE on average was most likely to be more costly but also more effective for improving QALYs than UC. The CEAC indicated that the probability of AE being cost-effective compared with UC was ≥35 to 81% depending on the WTP threshold. Online supplemental additional file 7, figure A3 shows CE planes and CEACs for 3 months follow-up.

The multiple one-way sensitivity analyses illustrated that altering costs 20% below and above estimates used in the main analysis did not affect the ICERs significantly (figure 2). SE and AE were still considered to be cost-effective compared with UC (total costs, 0–12 months). Altering QALYs, however, had a significant effect on the ICERs and led to some ICERs above a WTP threshold of €27 500 per QALY gained.

Figure 2

Incremental cost-effectiveness ratio (ICER) tornado diagram for multiple one-way sensitivity analyses of (A) strength exercise (SE) versus usual care (UC) and (B) aerobic exercise (AE) versus usual care. The tail of each bar indicates the upper (light grey) and lower (dark grey) bound of the ICER when relevant costs/QALYs is varied 20% below and above estimates used in the main analysis (total costs, 0–12 months). The dashed line represents the willingness-to-pay (WTP) threshold to provide a reference for the ICERs. QALYs, quality-adjusted life-years.

Discussion

The present study evaluated the 1-year CE of 12 weeks of SE or AE compared with UC for patients with symptomatic knee OA. From a societal and healthcare perspective, both SE and AE were found to be cost-effective at a WTP threshold of €27 500. Three out of four calculations at 1 year resulted in negative ICERs, indicating cost-saving interventions. Whereas AE from a healthcare perspective resulted in an ICER of €8725 per QALY gained. Although the results were sensitive to different estimates of QALYs (due to very small group differences circling around zero), all ICERs were well below the WTP threshold of €27 500. From a societal perspective, the probability of SE or AE being cost-effective was ≥98% for all WTP thresholds. From a healthcare perspective, the probability of SE being cost-effective was ≥95% for all WTP thresholds, whereas the probability of AE being cost-effective varied between 35% and 81%, depending on the WTP threshold.

Direct comparability of this study with other studies is limited due to no similar study has been conducted within the Norwegian healthcare system. Furthermore, there is a great heterogeneity in the method used among CE studies on knee OA.15 Nevertheless, our findings are generally in accordance with previous research from other countries. In a recent systematic review,15 overall, exercise interventions with or without education and diet as adjunctive therapies for hip and/or knee OA were found to be cost-effective at conventional WTP thresholds in numerous healthcare systems. Yet, CE seems to depend on (among other things) the comparator. 15 out of 16 studies included in the systematic review indicated that exercise interventions (4 with education and 2 with diet) were cost saving or cost-effective compared with education or physician-delivered UC, whereas three studies indicated that exercise interventions were not cost-effective compared with physiotherapist-delivered UC.15 Of studies reporting cost-saving or CE, ICERs ranged from cost-saving to US$883 per QALY gained (2019 US dollar).15 Moreover, three studies from the Nordic countries35–37 published after the systematic review of Mazzei et al15 also reported that exercise interventions with or without education and diet as adjunctive therapies for hip and/or knee OA is cost-effective.

With estimated annular costs related to lower-limb OA as high as €817 billion in Europe alone5 and an expected increase in the future burden of the disease,1 2 cost-effective treatment strategies is highly needed.6 11 15 Our study adds to the body of evidence that implementation of structured exercise programmes in clinical practice is a cost-effective alternative to UC in a 1-year horizon.

The main limitation of the present study is that it was powered on the primary outcome of the three-arm RCT, not on the CE outcome. However, this is common practice in trial-based economic evaluations, due to the right-skewed distribution of cost data and the fact that an extremely large sample size is required to detect relevant cost differences, which in turn can be both infeasible and/or unethical.13 To address this limitation, as recommended, bootstrapped estimates were used for result interpretation.13 A second limitation of the present study is that we had missing data on variables (healthcare utilisation, productivity loss, EQ-5D-5L) used to estimate the outcome variables and had to replace missing values. This might have introduced some risk of bias. Yet, there were only minor differences between responders and non-responders at baseline, an appropriate method of imputation (multiple imputation)38 was used, and sensitivity analyses showed similar results when comparing complete-case analysis to the main analysis. A third limitation is the fact that we expect to have somewhat underestimated total healthcare utilisation and related costs. Self-reports tend to underestimate the true value of healthcare utilisation due to potential recall bias.39–42 Moreover, we lack data on medication use and diagnostic imaging. Nevertheless, we consider the impact of this to be of only minor importance in this study, as healthcare utilisation and related costs were measured equally in the three groups being compared. Furthermore, the majority of costs in this study were related to productivity loss; costs due to healthcare utilisation had only a minor impact. A fourth potential limitation is that productivity loss was self-reported and not measured with a standardised validated method. However, a recent meta-analysis supports satisfactory agreement between self-reported and registry data on the occurrence and duration of absenteeism.43 Also, we lack data on productivity loss related to reduced productivity while at paid work (presenteeism) and unpaid work, thus also expect to have somewhat underestimate the true value of productivity loss, especially among people not in paid work. Yet, we also consider the impact of this to be of only minor importance in this study, as productivity loss and related costs were measured equally in the three groups being compared. A fifth limitation is the lack of data on eligible participants that declined to participate or for other reasons were not invited. Due to limited resources and practical reasons related to recruitment from a broad network of clinicians, it was not possible to record information on all eligible participants during the data collection period. Thus, the risk of selection bias is present.

The main strength of the present study is that it was a CE study of a three-arm RCT evaluating the effect of two structured exercise programmes with relatively few exercises involved and high adherence,16 that it was conducted in line with the Panel on Cost-Effectiveness in Health and Medicine guidelines,19 20 preplanned with a published statistical analysis plan and reported in line with the CHEERS Statement.23 Also, bootstrapping methods were used to determine uncertainty around the ICERs, which is recommended to handle skewed cost data.44

Conclusion

In conclusion, the present study demonstrated that implementing SE or AE for patients with symptomatic knee OA is a cost-effective alternative compared with UC within the Norwegian healthcare system, at a WTP threshold of €27 500. Both are from a societal and healthcare perspective. Yet, the results were sensitive to different estimates of QALYs (due to very small group differences). Moreover, the result with regard to AE evaluated from a healthcare perspective should be interpreted with caution, due to some uncertainty surrounding the calculated ICER. Further studies are needed to confirm these findings. However, the present study gives further support to the notion that structured exercise programmes for patients with symptomatic knee OA have the potential to reduce the global burden of knee OA.

Data availability statement

Data are available on reasonable request. Data supporting findings of this study are not public available as participants have not consented for their data to be public available. However, data are available from the corresponding author on reasonable request and with permission of an ethics committee and the Oslo University Hospital (contact through corresponding author).

Ethics statements

Patient consent for publication

Ethics approval

The study protocol has been approved by the Norwegian Regional Committee for Medical Research Ethics (ref no. REK 10/223) and approved by the Data Inspectorate at Oslo University Hospital, Norway (ref no. 2012/334). The study was conducted in accordance with the Declaration of Helsinki, and all data were stored on a secure IT-platform Oslo University Hospital.

Acknowledgments

We would like to thank all study participants, Asbjørn Årøen, Jan Harald Røtterud, Nina Østerås, Even Jarstad, Kristin Bølstad, Marte Lund and Zina Zolic-Karlsson for their significant contribution. We thank all collaborating physiotherapy institutes, and the department of Radiology at Oslo University Hospital by Kristin Dehli, Kathrine Lamark and Ragnhild Gunderson. We thank the Oslo University Hospital, Lovisenberg Hospital and Akershus University Hospital for their assistance in recruiting participants.

References

Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

Footnotes

  • Contributors BEØ, MAR and MG designed the study. RMK, BEØ, MAR and MG contributed to the funding of the study. BEØ collected data for the study. RMK analysed the data. RMK, BEØ, MAR, EM and MG contributed to the interpretation of data. RMK drafted the manuscript with all authors contributing to reading, commenting and approving the final manuscript. RMK is resonsible for the overall content as guarantor.

  • Funding This study was supported by the Norwegian Fund for Post-Graduate Training in Physiotherapy (grant no. 195521 and 62486) and the Research Council of Norway (grant no. H10/213335). The open access fee was funded by Oslo Metropolitan University.

  • Disclaimer Funding organisations had no part in the planning, performing, or reporting of the study.

  • Competing interests None declared.

  • Patient and public involvement Patients and/or the public were involved in the design, or conduct, or reporting, or dissemination plans of this research. Refer to the Methods section for further details.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.